Abstract
Recently, a physiologically-based, segregated flow model that incorporates separate intestinal tissue and flow to both a nonabsorptive and an absorptive outermost layer (enterocytes) was shown to better describe the observations on route-dependent morphine glucuronidation in the rat small intestine than a traditional physiologically-based model. These theoretical models were expanded, as the segmental segregated flow model and the segmental traditional model, to view the intestine as three segments of equal lengths receiving equal flows to accommodate heterogeneities in segmental transporter and metabolic functions. The influence of heterogeneity in absorptive, exsorptive, and metabolic functions on drug clearance, bioavailability (F), and metabolite formation after intravenous and oral dosing was examined for the intestine when the tissue was the only organ of removal. Simulations were performed for first-order conditions, when drug partitioned readily (flow-limited distribution) or less readily (membrane-limited distribution) into intestinal tissue, and for different gastrointestinal transit times. The intestinal clearance was found to be inversely related to the rate constant for absorption of a drug that was subjected to secretion and was positively correlated with the metabolic and secretory intrinsic clearances. F was positively correlated with the absorption rate constant but was inversely related to the metabolic and secretory intrinsic clearances. The gastrointestinal transit time decreased metabolite formation, increased clearance, and decreasedF. The simulations further showed that a descending metabolic intrinsic clearance yielded a lower F and an ascending segmental distribution of metabolic intrinsic clearance yielded a higher F.
The small intestine is endowed with transporters that effect the penetration of drugs across the luminal (or apical) membrane into the cell against a concentration gradient (for review, see Tsuji and Tamai, 1996; Lin et al., 1999). Permeation via passive diffusion of lipophilic drugs exists and is highly correlated to the surface area of contact and the pKa that influence the degree of ionization and hence lipophilicity. The varying abundance of the villi along the intestinal length constitutes differing surface areas among the intestinal segments, being highest at the duodenum and upper jejunum and lowest toward the ileum (Magee and Dalley, 1986). Net apical to basolateral transport is additionally influenced by the presence of drug binding, metabolizing enzymes, transporters for efflux and basolateral transport, and the gastrointestinal motility that modulates drug transit time.
Several models have been developed to describe processes of intestinal absorption, metabolism, and secretion simultaneously (Yu and Amidon, 1998; Ito et al., 1999; Cong et al., 2000). A traditional, physiologically-based model (TM2), which regards the intestine as a single homogeneous compartment with all of the intestinal blood flow perfusing the tissue, has been developed to account for oral drug bioavailability (Doherty and Pang, 2000). However, this and other existing models fail to predict route-dependent intestinal metabolism, namely little metabolism occurs after systemic dosing, but notable metabolism exists following oral dosing. This led to the development of the segregated flow model (SFM) (Cong et al., 2000) that describes the majority of the intestinal blood flow to the nonabsorptive and nonmetabolizing serosal and submucosa regions, and only partial flow (Granger et al., 1980) to the absorptive and metabolizing, enterocyte region at the villus tips of the mucosa where the metabolic enzymes and the P-glycoprotein reside. The SFM was shown able to describe a greater extent of intestinal metabolism with oral over systemic dosing or the route-dependent intestinal hydrolysis of enalapril (Pang et al., 1985) and (−)-6-aminocarbovir (Wen et al., 1999), glucuronidation of morphine (Doherty and Pang, 2000), and oxidation of midazolam (Paine et al., 1996, 1997; Thummel et al., 1996).
However, variation in segmental absorption was shown to exist for oxyprenolol (Godbillon et al., 1987), talinolol (Grámatte et al., 1996), amoxicillin (Barr et al., 1994), lefradafiban (Drewe et al., 2000), allopurinol (Patel and Kramer, 1986), thymidine analogs (Park and Mitra, 1992), and benzoate (Cong et al., 2001) in both human and animal studies. Undoubtedly, heterogeneity of absorptive carriers would bring about variations in absorption. Regional or segmental distribution of apical transporters—the oligopeptide transporter, PEPT1 (Fei et al., 1994), the apical bile salt transporter (Shneider et al., 1995; Aldini et al., 1996), the organic anion transporting polypeptide 3 (Walters et al., 2000), the monocarboxylic acid transporter 1 (Tamai et al., 1999; Cong et al., 2001), and the nucleoside transporter (Ngo et al., 2001)—is well recognized. Heterogeneity is further known to exist for both metabolic enzymes and efflux transporters. The cytochrome P450 3A (Hoensch et al., 1976; Bonkovsky et al., 1985; Paine et al., 1996,1997; Thummel et al., 1996; Lown et al., 1997; Li et al., 2002), sulfotransferases, glutathione S-transferases and the UDP-glucuronosyltransferases (Clifton and Kaplowitz, 1977; Pinkus et al., 1977; Schwarz and Schwenk, 1984; Koster et al., 1985; Coles et al., 2002) are higher at the proximal end than the distal intestine. The multidrug resistance-associated protein 2 for intestinal exsorption follows the distribution of the cytochrome P450s and conjugation enzymes (Gotoh et al., 2000; Mottino et al., 2000) whereas the MDR1 gene product, the 170 kD P-glycoprotein (Lown et al., 1997;Collett et al., 1999; Nakayama et al., 2000; Stephens et al., 2001) is higher in the jejunum/ileum than other parts of the intestine. The basolateral Mrp3, in contrast to multidrug resistance-associated protein 2 that is higher in jejunum and duodenum, is more prevalent in the ileum and colon (Rost et al., 2002). The manner in which heterogeneity impacts drug bioavailability is virtually unknown.
In this communication, we examined, in theoretical models, the influence of apical transporter and cellular metabolic heterogeneities, drug partitioning, and gastrointestinal transit on the area under the concentration-time curves after oral (AUCpo) and intravenous (AUCiv) dosing for estimation of the clearance (CLiv) and bioavailability (F) for a compound that is removed solely by the intestine by secretion and metabolism. We further tested the hypothesis that heterogeneity in transporter and metabolic functions, together with gastrointestinal transit and drug partitioning properties, affect the extent of metabolite formation after oral and intravenous drug dosing.
Materials and Methods
Physiologically Based Intestinal Models.
The schematic depictions of the segmental traditional model (STM; Fig.1) and the segmental segregated flow model (SSFM; Fig. 2) are shown. These theoretical models are extensions of the published TM and the SFM (Cong et al., 2000) that only contained nonsegmented compartments. The two previous models are physiologically-based models, and the difference between the models is division of the intestinal tissue into the serosal (nonabsorptive and nonmetabolizing) and enterocyte (absorptive and metabolizing) regions and partitioning of flow for the SFM. With only a fraction (fQ) of the total intestinal flow (Qint) perfusing the enterocyte region, less drug is exposed to metabolic enzymes with intravenous administration.
An expansion of these schemes yields the STM and the SSFM, in which the intestinal tissue is divided into three segments of equal lengths, for the sake of simplification (Figs. 1 and 2). Again, both are physiologically-based models established to accommodate heterogeneous intestinal processes among the segmental regions. The theory is based on a central or reservoir compartment from which clearance exists for other parallel organs, CLo, and an eliminating intestine compartment, with its blood and lumen compartments. Common features exist between the models. Drug partitioning between intestinal tissue (serosal and enterocyte regions) and intestinal blood is described by clearances, CLd1 and CLd2, respectively, as depicted. The volumes of the lumen for each segment are identical, namely,Vlum1 =Vlum2 =Vlum3 and equals 1/3 ofVlum, the total volume of the lumen. Metabolism exists only in the cell compartment with metabolic intrinsic clearance, CLint,mi where subscript “i” denotes segment 1, 2, or 3 for the formation of metaboliteMint1,Mint2, andMint3, respectively. Permeation of drug from the lumen into the intestinal tissue, whether mediated by carriers (Tsuji and Tamai, 1996) or passive diffusion, is associated with the absorption rate constant, kaiin which subscript “i” denotes segment 1, 2, or 3. Secretion from intestine tissue or enterocyte layer back into the lumen occurs with the intrinsic clearance, CLint,seci, where subscript “i” denotes segment 1, 2, or 3. Analogously, movement of drug along the intestinal lumen of segments 1, 2, and 3 is denoted as CLGIT1, CLGIT2, and CLGIT3, respectively. The total volume of the intestinal tissue (Vint) and intestinal blood (Vintb) for STM is the sum of those of the three segments, whereVint1 =Vint2 =Vint3 = 1/3Vint andVintb1 =Vintb2 =Vintb3 = 1/3Vintb. Each segment receives 1/3 of the intestinal flow or Qint/3 for the STM (Fig. 1). For the SSFM, the total intestinal tissue volume,Vint, is the sum of the serosal (Vs) and enterocyte (Ven) volumes, whereVs1 =Vs2 =Vs3 = 1/3Vs, andVen1 =Ven2 =Ven3 = 1/3Ven. The corresponding intestinal blood volume is further divided as the serosal blood (sb) and enterocyte blood (enb) volumes, such thatVsb1 =Vsb2 =Vsb3 = 1/3Vsb, andVenb1 =Venb2 =Venb3 = 1/3Venb. The segmental flow for the enterocyte region isfQQint/3 and that for the serosal region is (1 −fQ)Qint/3. Metabolism takes place in the enterocyte compartments (Fig. 2).
Simulations.
For the present models, the unbound fraction is assumed to be unity, and linear or first-order conditions prevail. The mass balance equations for the above schemes for the STM (eqs. 1 to 15,) and SSFM (eqs. 16 to 36) were written for simulation of data. Simulation was performed with volumes and flow rates shown in Table 1. The values have been chosen previously to describe glucuronidation of morphine and benzoate absorption in the vascularly perfused, recirculating rat small intestine preparation (Doherty and Pang, 2000; Cong et al., 2001). For values pertaining to the simulation on other species, the volume and flow values will need to be scaled-up or scaled-down. For the SSFM, the segmental enterocyte region was assumed to receive 10% of the intestinal flow to that segment (or fQQint/3, in whichfQ = 0.1), whereas the segmental serosal region receives 90% of the intestinal flow to the segment [(1− fQ)Qint/3].
Two strategies were undertaken. The first strategy took the view that all transporters for absorption (kai = 3 min−1) and exsorption (CLint,seci = 1 ml/min) and the metabolic enzymes (CLint,mi = 0.1 ml/min) were evenly dispersed among the segments for both STM and SSFM. Values of 10 ml/min (flow-limited transport, no transmembrane barrier) and 0.9 ml/min (membrane-limited transport) were assigned to CLd1 and CLd2, whereas values of 0.1 ml/min (faster gastrointestinal transit time) and 0.01 ml/min (slower gastrointestinal transit time) were used for CLGITi. To assess the influence of the absorption rate constant, the metabolic and secretory intrinsic clearances on AUC and metabolite formation, kai was varied between 0.01 to 5 min−1, CLint,mi was varied from 0.01 to 5 ml/min, and CLint,seci was varied from 0.1 to 1, 5, and 20 ml/min. The second strategy was to vary the total amount of absorptive, exsorptive, and metabolic activities for the intestine (sum of all ith segments) differentially among the segments (see Tables 2 to 5) to explore the effect of heterogeneity.
Simulation was performed with the program, Scientist (Micromath Scientific Software, Salt Lake City, UT). Recovery amounted to 100% at all times of simulation. The concentration versus time data for drug after intravenous (administration into reservoir) and oral (administration into lumen) were simulated to a time until drug concentration in reservoir became zero and the cumulative amount of metabolite became constant. The drug data were used to calculate the AUC by the trapezoidal rule and for estimation of the clearance (CLiv) and bioavailability (F).
Results
Homogeneous Distributions—Effect of kaiand CLint,seci on CLiv, F, and Metabolite Formation at CLint,mi of 0.1 ml/min.
Contrary to previous theories on the lack of influence of the absorption rate constant on the intestinal clearance, CLiv bore an inverse relation to thekai and was influenced positively by the secretory intrinsic clearance, CLint,seci(Fig. 3) for both STM (A and B) and SSFM (C and D). A higher CLGITi brought about higher CLiv for both STM and SSFM. When drug displays rapid partitioning (CLd1 = CLd2 = 10 ml/min; Fig. 3, B and D), values of CLiv were much higher than comparable values at CLd1 and CLd2 of 0.9 ml/min (cf. Fig. 3, B versus A and D versus C). Bioavailability, F, increased with increases in kai but decreased with CLint,seci (Fig.4), and high CLGITiled to lower F for both models. Higher values ofF were observed for the STM versus the SSFM (cf. Fig. 4, A and B versus C and D). For drug displaying rapid partitioning (CLd1 = CLd2 = 10 ml/min, Fig. 4, B and D), values of F were higher than comparable values at CLd1 and CLd2 of 0.9 ml/min (cf. Fig. 4, B versus A and D versus C). It was interesting to note that higher values for the kaitended to diffuse the effects of secretion (CLint,seci) on F. The patterns of metabolite formation were similar for the STM and SSFM. The extents of metabolite formation for both values of CLd1 and CLd2 (0.09 and 10 ml/min) were generally similar and were lower with higher values of CLGITi. Metabolite formation decreased with CLint,seci, and the amount of metabolite formed was greater for oral than for intravenous administration (cf. Fig. 5, A and B for STM and C and D for SSFM).
Homogeneous Distributions—Effect of CLint,mi and CLint,seci on CLiv, F, and Metabolite Formation at kai of 3 ml/min.
CLiv rose positively with higher values of CLint,mi, CLint,seci, and CLGITi (Fig. 6) for both STM (A and B) and SSFM (C and D). Higher values of CLGITi, CLd1, and CLd2 brought about higher CLiv for both STM and SSFM (cf. Fig. 6, B versus A for STM and D versus C for SSFM). Bioavailability, F, decreased with increasing values of the metabolic intrinsic clearance and CLGITi for both models (Fig.7). Values of F were generally higher for the STM in relation to those for the SSFM (cf. A and B versus C and D). The patterns of metabolite formation were similar for the STM and SSFM and were similar for both CLd1and CLd2 values chosen. Metabolite formation for oral dosing exceeded that for intravenous administration (Fig.8, compare A and B for STM and C and D for SSFM), and was decreased by CLint,seci and CLGITi.
Heterogeneous Distributions of kai, CLd1, CLd2, CLGITi,CLint,mi, and CLint,seci on CLiv,F, and Metabolite Formation.
The distributions of the segmental intrinsic clearances for secretion and metabolism and the absorption rate constant strongly influenced CLiv, F,Mpo andMiv for the STM (Table 2 for CLd1 = CLd2 = 0.9 ml/min and Table 3 for CLd1 = CLd2 = 10 ml/min) and the SSFM (Table 4 for CLd1 = CLd2 = 0.9 ml/min and Table 5 for CLd1 = CLd2 = 10 ml/min). The CLiv and F values were generally lower for the SSFM compared with the STM, although the patterns on metabolite formation remained similar. The lowest F occurred for case 24 (Tables 2 to 5) when kaiand CLint,mi were decreasing along the length of the intestine, whereas the CLint,seci increased along the length of the intestine (Fig.9, left panel). Low values ofF persisted for other cases (19 to 27) in which CLint,mi was decreasing along the length of the intestine, even when kai and CLint,seci displayed varying distributions (homogeneous, increasing or decreasing gradient). The highestF was observed for case 16 (Tables 2 to 5) when the metabolic and secretory intrinsic clearances (ascending from segment 1 to segment 3) were staggered in an opposite configuration tokai (descending from segment 1 to segment 3) (Fig. 9, right panel). Similar patterns of Fprevailed for cases 10 to 18 in which CLint,midisplayed an increasing, segmental gradient, even whenkai and CLint,seci exhibited varying distributions (homogeneous, increasing or decreasing gradient). The Fvalues were intermediate when CLint,mi was homogeneously distributed among segmental regions (cases 1 to 9). These trends were similar at CLGITi of 0.01 ml/min (Tables 2 and 4) and at CLGITi of 0.1 ml/min (Tables 3 and 5). The F values, however, did not directly reflect CLiv, which was lowest for case 25 and highest for case 7 (see Tables 2 to 5). The same comment applied to metabolite formation after oral dosing; these values do not correlate with F, although case 24 consistently provided the lowest metabolite formation (% dose) after oral dosing.
Discussion
It is well recognized that drug absorption is the net result of complex interactions of apical to basolateral transport, metabolism and exsorption/efflux in the intestine for which much heterogeneity exists. In addition to the named segmental differences in surface area, flow, absorptive carriers, metabolic enzymes, and efflux transporters, bile acids are known to affect the permeability of intestinal mucosa, the solubilization of drug and/or suppression of thermodynamic activity after its involvement in the micellar complex (Emori et al., 1995). It is surmised that the influence of bile acid is greater at the sphincter of Oddi and the proximal duodenum. Hence, segmental differences in drug absorption are not unexpected outcomes.
To date, none of the proposed models (Yu and Amidon, 1998; Ito et al., 1999; Lin et al., 1999; Cong et al., 2000) is able to describe heterogeneous events in any cohesive fashion. The development of the STM and SSFM therefore provides the first theoretical models that provide predictive information on the impact of intestinal heterogeneities on drug absorption. Analogous to that shown previously for TM and SFM (Cong et al., 2000), differences exist between the STM and SSFM due to the segregation of flow for the segments of SSFM. However, common conclusions may be made for both models. Because of drug secretion, reabsorption with kaiaffects not only CLiv and F but also metabolite formation (Figs. 3 to 5). The absorption rate constant decreases CLiv but increases F and metabolite formation due to recycling of drug back to the circulation. The kai neutralizes some of the effects of drug secretion, and whenkai is large, secretion effects become minimal. The gastrointestinal transit time, denoted by CLGITi, effectively removes drug from the lumen and precludes absorption. Increases in CLGITibring about increases in CLiv but decreaseF, whereas rapid drug partitioning further brings about higher CLiv and F (Figs. 3 to 8; Tables 2 to 5). The metabolic and secretory intrinsic clearances affect CLiv and F for the STM to a much greater degree than for the SSFM (Figs. 3 and 4 and Figs. 6 and 7; cf. A and B versus C and D). Both metabolic and secretory intrinsic clearances increase CLiv but decreaseF.
The presence of secretion also affects metabolite formation, and a greater CLint,seci results in lowered metabolite formation. A similar observation was made for a theoretical examination on the time course of metabolite production in Caco-2 cells (D. Tam and K. S. Pang, unpublished data) even though metabolite formation eventually equals the dose in this stagnant system. By contrast, metabolite formation is reduced by rapid gastrointestinal transit. Metabolite formation for oral dosing is usually greater than that with intravenous dosing (Figs. 5 and 8 and Tables 2 to 5). However, whenkai is very low (<0.01 min−1) and CLGITi (>0.1 ml/min) is high, metabolite formation after intravenous dosing can exceed that for oral dosing (unpublished simulations).
Heterogeneity in metabolic intrinsic clearance among the intestinal segments strongly impacts F. Lower values of Fexisted when the gradient of metabolic activities was more proximal and dwindles toward the ileum (Fig. 9, left panel). Alternately, higherF values were observed when the segmental, metabolic intrinsic clearance was increasing toward the ileum (Fig. 9, right panel). These comments hold true regardless of the distributions (homogeneous, increasing or decreasing gradient) ofkai and CLint,seci. Metabolite formation was influenced by all of the heterogeneities, and no succinct pattern was readily identified.
The present simulation study with two theoretical models, the STM and SSFM, revealed possible interactions of the absorption rate constant, metabolic and secretory intrinsic clearances, flow, and gastrointestinal transit on drugs of varying tissue-partitioning characteristics. Although the present exploration of the theoretical models presented only a limited scope of the influence of the various segmental variables on drug bioavailability and metabolite formation, important observations nonetheless arose. The study showed that the absorption rate constant, in contrast to cases where secretion is nonexistent, strongly affected the clearance and bioavailability. Segmental metabolic heterogeneity was another important factor that influences CLiv and F.
Appendix
The mass-balanced rate equations for the STM are shown below.
For the change of drug concentrations (D) in the reservoir (compartment “R”)
For the change of drug (D) concentrations in the reservoir (compartment “R”)
Footnotes
-
↵1 Present Address: Amgen Inc., Small molecule Pharmacokinetics and Drug Metabolism, Amgen Center, Thousand Oaks, CA 91320-1799
-
This work was supported by the Canadian Institute for Health Research, Grant MOP36457. Debbie Tam was a recipient of a Natural Sciences and Engineering Research Council summer studentship.
- Abbreviations used are::
- TM
- traditional, physiologically-based model
- SFM
- segregated flow model
- STM
- segmental traditional model
- SSFM
- segmental segregated flow model
- sb
- serosal blood
- enb
- enterocyte blood
- AUCiv and AUCpo
- areas under the curve for intravenous and oral dosing, respectively
- CLd1 and CLd2
- transfer clearances from blood to tissue compartment, and from tissue to blood compartment, respectively
- CLint,m and CLint,sec are the metabolic and secretory intrinsic clearances for the total intestine
- respectively
- CLint,mi and CLint,seci
- metabolic and secretory intrinsic clearances for the ith segment (i = 1, 2, or 3), respectively
- CLiv
- intravenous clearance
- CLGIT
- the gastrointestinal luminal transit clearance for the entire intestine
- CLGITi
- the gastrointestinal luminal transit clearance for the ith segment
- F
- bioavailability
- fQ
- fraction of the total intestinal flow perfusing the enterocyte region
- ka
- absorption rate constant for the intestine
- kai
- absorption rate constant for the ith segment
- Mint
- total amount of metabolite formed, with subscripts p.o. and i.v. further denoting oral and intravenous dosing, respectively
- Minti
- metabolite formation from the ith segment
- Qint
- total blood flow to the intestine
- Ven and Venb
- volumes of the enterocyte layer and the blood to the enterocyte layer, respectively
- Veni and Venbi
- volumes of the enterocyte layer and the blood to the enterocyte layer of the ith segment, respectively
- Vint and Vintb
- volumes of intestinal tissue and intestinal blood for the entire intestine
- Vinti and Vintbi
- volumes of intestinal tissue and intestinal blood for the ith segment
- Vs and Vsb
- volumes of the serosal layer and the blood to the serosal layer, respectively
- Vsi and Vsbi
- volumes of the serosal layer and the blood to the serosal layer of the ith segment, respectively
- Received August 28, 2002.
- Accepted December 17, 2002.
- The American Society for Pharmacology and Experimental Therapeutics